yamaa Review ask
As a clinical trial statistician or programmer, challenge anything in this benchmark: describe a problem in Comments, or propose a fix with the Edit button on any source file.
- Task: Would a real study need this derivation, specified this way? Does it follow CDISC ADaM: variable names, controlled terms, record structure? Flag any rule that is ambiguous or that a sponsor would handle differently.
- Inputs: Do they look like data collected in a real trial, and do they include the situations you would expect to meet in one?
- Expected output: Would independent QC programming reproduce every value from these inputs and rules?
The other tabs beside Comments are background for assessment, not under review.
Summary
EditUnder reviewWhat this benchmark meansREADME.mdGoal: derive TR01SDT, TR01EDT, TR02SDT, TR02EDT,
TRT01A, TRT02A, and WASHDUR to capture each treatment
period and the washout between them, with one record per
subject.
Input: demographics listing each subject, plus exposure
records carrying the epoch (EPOCH), the treatment
given (EXTRT), start and end dates (EXSTDTC, EXENDTC), and
a sequence number (EXSEQ).
Variables:
TR01SDTis the earliestEXSTDTCamong the subject's exposure records in theTREATMENT 1epoch and a recorded start date; empty when the subject has no such record.TR01EDTis the latestEXENDTCamong the subject's exposure records in theTREATMENT 1epoch and a recorded end date; empty when the subject has no such record.TR02SDTis the earliestEXSTDTCamong the subject's exposure records in theTREATMENT 2epoch and a recorded start date; empty when the subject has no such record.TR02EDTis the latestEXENDTCamong the subject's exposure records in theTREATMENT 2epoch and a recorded end date; empty when the subject has no such record.TRT01Ais theEXTRTfrom the subject's earliest exposure record in theTREATMENT 1epoch, earliest byEXSTDTCwith the lowerEXSEQbreaking ties on the same day and a record with no start date sorting last; eitherVITAMIN D3orPLACEBO, and empty when the subject has no period-one exposure.TRT02Ais theEXTRTfrom the subject's earliest exposure record in theTREATMENT 2epoch, chosen the same way; eitherVITAMIN D3orPLACEBO, and empty when the subject has no period-two exposure.WASHDURis the number of days strictly betweenTR01EDTandTR02SDT, counting neither endpoint, so periods that touch give zero; empty when either date is missing.
Note: every value comes only from exposure records in its
own period, so period-one dates and treatment never mix with
period-two records. A subject with no period-two exposure
leaves TR02SDT, TR02EDT, TRT02A, and WASHDUR empty
together.
Inputs
Under review2 source filesinput/dm.csv
Edit6 rows| STUDYID | USUBJID |
|---|---|
| YAMAA-01 | |
| YAMAA-01 | |
| YAMAA-01 | |
| YAMAA-01 | |
| YAMAA-01 | |
| YAMAA-01 |
input/ex.csv
Edit17 rows| STUDYID | USUBJID | EXSEQ | EPOCH | EXTRT | EXSTDTC | EXENDTC |
|---|---|---|---|---|---|---|
| YAMAA-01 | 1 | TREATMENT 1 | VITAMIN D3 | 2025-01-06 | 2025-01-16 | |
| YAMAA-01 | 2 | TREATMENT 1 | VITAMIN D3 | 2025-01-17 | 2025-01-26 | |
| YAMAA-01 | 3 | TREATMENT 2 | PLACEBO | 2025-02-10 | 2025-03-02 | |
| YAMAA-01 | 3 | TREATMENT 2 | PLACEBO | 2025-02-10 | 2025-03-02 | |
| YAMAA-01 | 1 | TREATMENT 1 | PLACEBO | 2025-01-08 | 2025-01-28 | |
| YAMAA-01 | 2 | TREATMENT 2 | VITAMIN D3 | 2025-02-12 | 2025-02-22 | |
| YAMAA-01 | 3 | TREATMENT 2 | VITAMIN D3 | 2025-02-23 | 2025-03-04 | |
| YAMAA-01 | 1 | TREATMENT 1 | VITAMIN D3 | 2025-01-10 | 2025-01-20 | |
| YAMAA-01 | 2 | TREATMENT 1 | VITAMIN D3 | 2025-03-10 | 2025-03-15 | |
| YAMAA-01 | 5 | TREATMENT 1 | PLACEBO | 2025-03-10 | 2025-03-20 | |
| YAMAA-01 | 1 | TREATMENT 2 | PLACEBO | 2025-04-01 | 2025-04-15 | |
| YAMAA-01 | 1 | TREATMENT 1 | PLACEBO | 2025-04-01 | 2025-04-10 | |
| YAMAA-01 | 2 | TREATMENT 2 | VITAMIN D3 | 2025-04-11 | 2025-04-25 | |
| YAMAA-01 | 1 | TREATMENT 1 | VITAMIN D3 | 2025-05-01 | 2025-05-10 | |
| YAMAA-01 | 2 | TREATMENT 1 | PLACEBO | — | 2025-05-12 | |
| YAMAA-01 | 3 | TREATMENT 2 | PLACEBO | 2025-06-01 | — | |
| YAMAA-01 | 4 | TREATMENT 2 | PLACEBO | 2025-06-02 | 2025-06-20 |
Expected output
Under review6 expected rowsexpected/adsl.csv
Edit6 rows| STUDYID | USUBJID | TR01SDT | TR01EDT | TR02SDT | TR02EDT | TRT01A | TRT02A | WASHDUR |
|---|---|---|---|---|---|---|---|---|
| YAMAA-01 | 2025-01-06 | 2025-01-26 | 2025-02-10 | 2025-03-02 | VITAMIN D3 | PLACEBO | 14 | |
| YAMAA-01 | 2025-01-08 | 2025-01-28 | 2025-02-12 | 2025-03-04 | PLACEBO | VITAMIN D3 | 14 | |
| YAMAA-01 | 2025-01-10 | 2025-01-20 | — | — | VITAMIN D3 | — | — | |
| YAMAA-01 | 2025-03-10 | 2025-03-20 | 2025-04-01 | 2025-04-15 | VITAMIN D3 | PLACEBO | 11 | |
| YAMAA-01 | 2025-04-01 | 2025-04-10 | 2025-04-11 | 2025-04-25 | PLACEBO | VITAMIN D3 | 0 | |
| YAMAA-01 | 2025-05-01 | 2025-05-12 | 2025-06-01 | 2025-06-20 | VITAMIN D3 | PLACEBO | 19 |
Shaded columns are derived— = missing value
Comments and assessment materials
Not under review
| Dataset Name | Variable Name | Variable Label | Type | Codelist/Controlled Terms | Core | Computational Method | Origin |
|---|---|---|---|---|---|---|---|
| ADSL | STUDYID | Study Identifier | character | Copy value from DM.STUDYID. | Predecessor | ||
| ADSL | USUBJID | Unique Subject Identifier | character | Copy value from DM.USUBJID. | Predecessor | ||
| ADSL | TR01SDT | Date of First Exposure in Period 01 | numeric | ISO 8601 | Aggregate MIN(EX.EXSTDTC) over records matching on the shared output keys, where EX.EPOCH = 'TREATMENT 1'. | Derived | |
| ADSL | TR01EDT | Date of Last Exposure in Period 01 | numeric | ISO 8601 | Aggregate MAX(EX.EXENDTC) over records matching on the shared output keys, where EX.EPOCH = 'TREATMENT 1'. | Derived | |
| ADSL | TR02SDT | Date of First Exposure in Period 02 | numeric | ISO 8601 | Aggregate MIN(EX.EXSTDTC) over records matching on the shared output keys, where EX.EPOCH = 'TREATMENT 2'. | Derived | |
| ADSL | TR02EDT | Date of Last Exposure in Period 02 | numeric | ISO 8601 | Aggregate MAX(EX.EXENDTC) over records matching on the shared output keys, where EX.EPOCH = 'TREATMENT 2'. | Derived | |
| ADSL | TRT01A | Actual Treatment for Period 01 | character | VITAMIN D3, PLACEBO | Copy value from EX.EXTRT where EX.EPOCH = 'TREATMENT 1', keeping the first ordered by EX.EXSTDTC then EX.EXSEQ. | Predecessor | |
| ADSL | TRT02A | Actual Treatment for Period 02 | character | VITAMIN D3, PLACEBO | Copy value from EX.EXTRT where EX.EPOCH = 'TREATMENT 2', keeping the first ordered by EX.EXSTDTC then EX.EXSEQ. | Predecessor | |
| ADSL | WASHDUR | Washout Duration (Days) | numeric | Whole days from TR01EDT to TR02SDT, counting neither endpoint. | Derived |
| Variable | Permitted values |
|---|---|
| TRT01A | VITAMIN D3, PLACEBO |
| TRT02A | VITAMIN D3, PLACEBO |
| Version | Date | Author | Description | Reviewer | Sign-off |
|---|---|---|---|---|---|
| 1.3 | 2026-09-29 | yamaa docs build | Show study codelist names, external dictionaries, and ISO 8601 formats in variable rows; list the document's codelist values and metadata | ||
| 1.2 | 2026-09-29 | yamaa docs build | Classify ADaM copies from a declared input as Predecessor and recodes as Derived, and SDTM recodes of locally derived columns as Derived | ||
| 1.1 | 2026-09-29 | yamaa docs build | Resolve lookups into the donor record they select, and give every derivation shape a plain-language rule instead of its raw YAML | ||
| 1.0 | yamaa docs build | Initial generation from spec.yaml |
prompts/adam-adsl-crossover-periods/full.md
Edit28 linesFollowing CDISC ADaM standards, use the provided DM and EX datasets to create an ADSL dataset with one record per subject.
The output dataset should contain the following columns in this order: STUDYID, USUBJID, TR01SDT, TR01EDT, TR02SDT, TR02EDT, TRT01A, TRT02A, WASHDUR
TR01SDT is the earliest recorded start date among the subject's exposure records in the TREATMENT 1 epoch; it has no value when the subject has no such record. TR01EDT is the latest recorded end date there. TR02SDT and TR02EDT are the same for the TREATMENT 2 epoch. Every value comes only from exposure records in its own period.
TRT01A is the treatment given on the subject's earliest exposure record in the TREATMENT 1 epoch, earliest by start date with the lower sequence number breaking ties on the same day and a record with no start date sorting last; either VITAMIN D3 or PLACEBO, and empty when the subject has no period-one exposure. TRT02A is chosen the same way in the TREATMENT 2 epoch.
WASHDUR is the number of days strictly between the end of period one and the start of period two, counting neither endpoint, so periods that touch give zero; it has no value when either date is missing. A subject with no period-two exposure leaves the period-two dates and treatment and the washout empty together.
Read the source datasets from /app/input and save the completed dataset as /app/output/adsl.csv.
prompts/adam-adsl-crossover-periods/conventions.md
Edit13 linesFollowing CDISC ADaM standards, use the provided DM and EX datasets to create an ADSL dataset with one record per subject.
The output dataset should contain the following columns in this order: STUDYID, USUBJID, TR01SDT, TR01EDT, TR02SDT, TR02EDT, TRT01A, TRT02A, WASHDUR
TRT01A and TRT02A are VITAMIN D3 or PLACEBO, or empty.
WASHDUR counts neither the end of period one nor the start of period two.
Read the source datasets from /app/input and save the completed dataset as /app/output/adsl.csv.
prompts/adam-adsl-crossover-periods/brief.md
Edit9 linesFollowing CDISC ADaM standards, use the provided DM and EX datasets to create an ADSL dataset with one record per subject.
The output dataset should contain the following columns in this order: STUDYID, USUBJID, TR01SDT, TR01EDT, TR02SDT, TR02EDT, TRT01A, TRT02A, WASHDUR
Read the source datasets from /app/input and save the completed dataset as /app/output/adsl.csv.
run.R
Edit4 lineslibrary(yamaa)adsl <- yamaa_domain("spec.yaml")$outputadslrun.py
Edit4 linesimport yamaaadsl = yamaa.yamaa_domain("spec.yaml").outputadslsolutions/adam-adsl-crossover-periods/result.R
Edit66 lines# Reference solution for the yamaa benchmark adam-adsl-crossover-periods (R track).## Written from the benchmark's full prompt and its inputs alone, as an agent# would write it. Harbor's oracle agent runs it to check that the prompt can# be solved and that the grader scores a correct answer 1.library(dplyr, warn.conflicts = FALSE)library(readr)dm <- read_csv( "/app/input/dm.csv", col_types = cols(.default = col_character()))ex <- read_csv( "/app/input/ex.csv", col_types = cols( EXSEQ = col_integer(), EXSTDTC = col_date(), EXENDTC = col_date(), .default = col_character() ))ex1 <- ex |> filter(EPOCH %in% "TREATMENT 1")ex2 <- ex |> filter(EPOCH %in% "TREATMENT 2")agg1 <- ex1 |> group_by(STUDYID, USUBJID) |> summarise( TR01SDT = if (all(is.na(EXSTDTC))) as.Date(NA) else min(EXSTDTC, na.rm = TRUE), TR01EDT = if (all(is.na(EXENDTC))) as.Date(NA) else max(EXENDTC, na.rm = TRUE), .groups = "drop" )agg2 <- ex2 |> group_by(STUDYID, USUBJID) |> summarise( TR02SDT = if (all(is.na(EXSTDTC))) as.Date(NA) else min(EXSTDTC, na.rm = TRUE), TR02EDT = if (all(is.na(EXENDTC))) as.Date(NA) else max(EXENDTC, na.rm = TRUE), .groups = "drop" )# The treatment on the earliest record in each period: earliest start date# with the lower sequence number breaking ties, records without a start# date sorting last.first1 <- ex1 |> arrange(STUDYID, USUBJID, is.na(EXSTDTC), EXSTDTC, EXSEQ) |> distinct(STUDYID, USUBJID, .keep_all = TRUE) |> transmute(STUDYID, USUBJID, TRT01A = EXTRT)first2 <- ex2 |> arrange(STUDYID, USUBJID, is.na(EXSTDTC), EXSTDTC, EXSEQ) |> distinct(STUDYID, USUBJID, .keep_all = TRUE) |> transmute(STUDYID, USUBJID, TRT02A = EXTRT)adsl <- dm |> left_join(agg1, by = c("STUDYID", "USUBJID")) |> left_join(agg2, by = c("STUDYID", "USUBJID")) |> left_join(first1, by = c("STUDYID", "USUBJID")) |> left_join(first2, by = c("STUDYID", "USUBJID")) |> mutate(WASHDUR = as.integer(TR02SDT - TR01EDT) - 1L) |> select( STUDYID, USUBJID, TR01SDT, TR01EDT, TR02SDT, TR02EDT, TRT01A, TRT02A, WASHDUR )dir.create("/app/output", showWarnings = FALSE)write_csv(adsl, "/app/output/adsl.csv", na = "")solutions/adam-adsl-crossover-periods/result.py
Edit65 lines# Reference solution for the yamaa benchmark adam-adsl-crossover-periods (Python track).## Written from the benchmark's full prompt and its inputs alone, as an agent# would write it. Harbor's oracle agent runs it to check that the prompt can# be solved and that the grader scores a correct answer 1.from pathlib import Pathimport polars as pldm = pl.read_csv("/app/input/dm.csv", infer_schema=False)ex = pl.read_csv("/app/input/ex.csv", infer_schema=False).with_columns( pl.col("EXSEQ").cast(pl.Int64, strict=False), pl.col("EXSTDTC").str.to_date(strict=False), pl.col("EXENDTC").str.to_date(strict=False),)ex1 = ex.filter(pl.col("EPOCH") == "TREATMENT 1")ex2 = ex.filter(pl.col("EPOCH") == "TREATMENT 2")agg1 = ex1.group_by(["STUDYID", "USUBJID"]).agg( TR01SDT=pl.col("EXSTDTC").min(), TR01EDT=pl.col("EXENDTC").max())agg2 = ex2.group_by(["STUDYID", "USUBJID"]).agg( TR02SDT=pl.col("EXSTDTC").min(), TR02EDT=pl.col("EXENDTC").max())# The treatment on the earliest record in each period: earliest start date# with the lower sequence number breaking ties, records without a start# date sorting last.first1 = ( ex1.sort(["EXSTDTC", "EXSEQ"], nulls_last=True) .unique(["STUDYID", "USUBJID"], keep="first", maintain_order=True) .select("STUDYID", "USUBJID", TRT01A="EXTRT"))first2 = ( ex2.sort(["EXSTDTC", "EXSEQ"], nulls_last=True) .unique(["STUDYID", "USUBJID"], keep="first", maintain_order=True) .select("STUDYID", "USUBJID", TRT02A="EXTRT"))adsl = ( dm.join(agg1, on=["STUDYID", "USUBJID"], how="left", maintain_order="left") .join(agg2, on=["STUDYID", "USUBJID"], how="left", maintain_order="left") .join(first1, on=["STUDYID", "USUBJID"], how="left", maintain_order="left") .join(first2, on=["STUDYID", "USUBJID"], how="left", maintain_order="left") # Days strictly between the periods, counting neither endpoint. .with_columns( WASHDUR=(pl.col("TR02SDT") - pl.col("TR01EDT")).dt.total_days() - 1 ) .select( "STUDYID", "USUBJID", "TR01SDT", "TR01EDT", "TR02SDT", "TR02EDT", "TRT01A", "TRT02A", "WASHDUR", ))Path("/app/output").mkdir(exist_ok=True)adsl.write_csv("/app/output/adsl.csv")solutions/adam-adsl-crossover-periods/result.sas
Edit36 lines/* opensas reference solution for the yamaa benchmark adam-adsl-crossover-periods. Reads the benchmark inputs and derives the requested output. Execute with the existing opensas runtime: opensas /app/output/result.sas. */data dm; length STUDYID USUBJID $1024; infile "/app/input/dm.csv" dsd dlm="," firstobs=2 truncover lrecl=32767; input STUDYID : $char1024. USUBJID : $char1024.;run;data ex; length STUDYID USUBJID EPOCH EXTRT EXSTDTC EXENDTC $1024; infile "/app/input/ex.csv" dsd dlm="," firstobs=2 truncover lrecl=32767; input STUDYID : $char1024. USUBJID : $char1024. EXSEQ EPOCH : $char1024. EXTRT : $char1024. EXSTDTC : $char1024. EXENDTC : $char1024.;run;data exposure; set ex; NODATE=missing(EXSTDTC); run;proc sort data=exposure; by STUDYID USUBJID EPOCH NODATE EXSTDTC EXSEQ; run;data firstex; set exposure; by STUDYID USUBJID EPOCH; if first.EPOCH; run;proc sql;create table starts1 as select STUDYID,USUBJID,min(EXSTDTC) as TR01SDT from ex where EPOCH='TREATMENT 1' and not missing(EXSTDTC) group by STUDYID,USUBJID;create table ends1 as select STUDYID,USUBJID,max(EXENDTC) as TR01EDT from ex where EPOCH='TREATMENT 1' group by STUDYID,USUBJID;create table starts2 as select STUDYID,USUBJID,min(EXSTDTC) as TR02SDT from ex where EPOCH='TREATMENT 2' and not missing(EXSTDTC) group by STUDYID,USUBJID;create table ends2 as select STUDYID,USUBJID,max(EXENDTC) as TR02EDT from ex where EPOCH='TREATMENT 2' group by STUDYID,USUBJID;create table joined as select a.*,b.TR01SDT,c.TR01EDT,d.TR02SDT,e.TR02EDT,f.EXTRT as TRT01A,g.EXTRT as TRT02A from dm aleft join starts1 b on a.STUDYID=b.STUDYID and a.USUBJID=b.USUBJIDleft join ends1 c on a.STUDYID=c.STUDYID and a.USUBJID=c.USUBJIDleft join starts2 d on a.STUDYID=d.STUDYID and a.USUBJID=d.USUBJIDleft join ends2 e on a.STUDYID=e.STUDYID and a.USUBJID=e.USUBJIDleft join firstex f on a.STUDYID=f.STUDYID and a.USUBJID=f.USUBJID and f.EPOCH='TREATMENT 1'left join firstex g on a.STUDYID=g.STUDYID and a.USUBJID=g.USUBJID and g.EPOCH='TREATMENT 2'; quit;data result; set joined; WASHDUR=.; if not missing(TR01EDT) and not missing(TR02SDT) then WASHDUR=max(0,input(TR02SDT,yymmdd10.)-input(TR01EDT,yymmdd10.)-1); run;proc sql; create table final as select STUDYID, USUBJID, TR01SDT, TR01EDT, TR02SDT, TR02EDT, TRT01A, TRT02A, WASHDUR from result;quit;proc export data=final outfile="/app/output/adsl.csv" dbms=csv replace; run;
Comments
Signed in with GitHub; saved in repository Discussions